On Wed, Jul 11, 2012 at 6:18 PM, Eliezer Croitoru <eliezer / ngtech.co.il> wrote:
> i indeed looked at it.
> it seems to fit my cache needs but not the filtering.
> it really simplified many things for me with the "expire"(setex) method.
> i need to store maximum of 3-6 minutes the key and the url so it fit's like
> hell to me.

Glad to hear that.

> for a more robust lookups i need DB and i can cache the results in redis
> because it's a one way ticket per domain for a long time.
> i can add a domain result for half a day and it will do a really good work.
>
> my main problem is that i am using forks and redis ruby library dont like
> it.
> and also my eventmachine server has a "clenup" nasty method that releases
> any old object when ending session so i do have a problem with it.
>
> if you have an idea on how to handle REDIS with forks i will be so happy.

I have no experience with the redis ruby library (I've only used it in
Java), but I guess the problem is that the library keeps a connection
open, and when you fork, reusing this connection across processes is a
problem. I don't know how this can be solved. If you are using this
one https://github.com/redis/redis-rb, it says to seek for help here:
"You can also ask for help at #redis-rb on Freenode."

Maybe you could try to open the connection in each process
individually after the fork.

Jesus.

Jesus.